1 /*
2  * Copyright (c) 2020 Ori Bernstein
3  * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 
18 #include <sys/types.h>
19 
20 #include <assert.h>
21 #include <errno.h>
22 #include <limits.h>
23 #include <stdint.h>
24 #include <stdio.h>
25 #include <stdlib.h>
26 #include <string.h>
27 
28 #include "got_error.h"
29 
30 #include "got_lib_deltify.h"
31 
32 #ifndef MIN
33 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
34 #endif
35 
36 /*
37  * The algorihm used here is FastCDC (Fast Content-Defined Chunking)
38  * https://www.usenix.org/conference/atc16/technical-sessions/presentation/xia
39  */
40 
41 static uint32_t geartab[256] = {
42     0x67ed26b7, 0x32da500c, 0x53d0fee0, 0xce387dc7, 0xcd406d90, 0x2e83a4d4,
43     0x9fc9a38d, 0xb67259dc, 0xca6b1722, 0x6d2ea08c, 0x235cea2e, 0x3149bb5f,
44     0x1beda787, 0x2a6b77d5, 0x2f22d9ac, 0x91fc0544, 0xe413acfa, 0x5a30ff7a,
45     0xad6fdde0, 0x444fd0f5, 0x7ad87864, 0x58c5ff05, 0x8d2ec336, 0x2371f853,
46     0x550f8572, 0x6aa448dd, 0x7c9ddbcf, 0x95221e14, 0x2a82ec33, 0xcbec5a78,
47     0xc6795a0d, 0x243995b7, 0x1c909a2f, 0x4fded51c, 0x635d334b, 0x0e2b9999,
48     0x2702968d, 0x856de1d5, 0x3325d60e, 0xeb6a7502, 0xec2a9844, 0x0905835a,
49     0xa1820375, 0xa4be5cab, 0x96a6c058, 0x2c2ccd70, 0xba40fce3, 0xd794c46b,
50     0x8fbae83e, 0xc3aa7899, 0x3d3ff8ed, 0xa0d42b5b, 0x571c0c97, 0xd2811516,
51     0xf7e7b96c, 0x4fd2fcbd, 0xe2fdec94, 0x282cc436, 0x78e8e95c, 0x80a3b613,
52     0xcfbee20c, 0xd4a32d1c, 0x2a12ff13, 0x6af82936, 0xe5630258, 0x8efa6a98,
53     0x294fb2d1, 0xdeb57086, 0x5f0fddb3, 0xeceda7ce, 0x4c87305f, 0x3a6d3307,
54     0xe22d2942, 0x9d060217, 0x1e42ed02, 0xb6f63b52, 0x4367f39f, 0x055cf262,
55     0x03a461b2, 0x5ef9e382, 0x386bc03a, 0x2a1e79c7, 0xf1a0058b, 0xd4d2dea9,
56     0x56baf37d, 0x5daff6cc, 0xf03a951d, 0xaef7de45, 0xa8f4581e, 0x3960b555,
57     0xffbfff6d, 0xbe702a23, 0x8f5b6d6f, 0x061739fb, 0x98696f47, 0x3fd596d4,
58     0x151eac6b, 0xa9fcc4f5, 0x69181a12, 0x3ac5a107, 0xb5198fe7, 0x96bcb1da,
59     0x1b5ddf8e, 0xc757d650, 0x65865c3a, 0x8fc0a41a, 0x87435536, 0x99eda6f2,
60     0x41874794, 0x29cff4e8, 0xb70efd9a, 0x3103f6e7, 0x84d2453b, 0x15a450bd,
61     0x74f49af1, 0x60f664b1, 0xa1c86935, 0xfdafbce1, 0xe36353e3, 0x5d9ba739,
62     0xbc0559ba, 0x708b0054, 0xd41d808c, 0xb2f31723, 0x9027c41f, 0xf136d165,
63     0xb5374b12, 0x9420a6ac, 0x273958b6, 0xe6c2fad0, 0xebdc1f21, 0xfb33af8b,
64     0xc71c25cd, 0xe9a2d8e5, 0xbeb38a50, 0xbceb7cc2, 0x4e4e73f0, 0xcd6c251d,
65     0xde4c032c, 0x4b04ac30, 0x725b8b21, 0x4eb8c33b, 0x20d07b75, 0x0567aa63,
66     0xb56b2bb7, 0xc1f5fd3a, 0xcafd35ca, 0x470dd4da, 0xfe4f94cd, 0xfb8de424,
67     0xe8dbcf40, 0xfe50a37a, 0x62db5b5d, 0xf32f4ab6, 0x2c4a8a51, 0x18473dc0,
68     0xfe0cbb6e, 0xfe399efd, 0xdf34ecc9, 0x6ccd5055, 0x46097073, 0x139135c2,
69     0x721c76f6, 0x1c6a94b4, 0x6eee014d, 0x8a508e02, 0x3da538f5, 0x280d394f,
70     0x5248a0c4, 0x3ce94c6c, 0x9a71ad3a, 0x8493dd05, 0xe43f0ab6, 0x18e4ed42,
71     0x6c5c0e09, 0x42b06ec9, 0x8d330343, 0xa45b6f59, 0x2a573c0c, 0xd7fd3de6,
72     0xeedeab68, 0x5c84dafc, 0xbbd1b1a8, 0xa3ce1ad1, 0x85b70bed, 0xb6add07f,
73     0xa531309c, 0x8f8ab852, 0x564de332, 0xeac9ed0c, 0x73da402c, 0x3ec52761,
74     0x43af2f4d, 0xd6ff45c8, 0x4c367462, 0xd553bd6a, 0x44724855, 0x3b2aa728,
75     0x56e5eb65, 0xeaf16173, 0x33fa42ff, 0xd714bb5d, 0xfbd0a3b9, 0xaf517134,
76     0x9416c8cd, 0x534cf94f, 0x548947c2, 0x34193569, 0x32f4389a, 0xfe7028bc,
77     0xed73b1ed, 0x9db95770, 0x468e3922, 0x0440c3cd, 0x60059a62, 0x33504562,
78     0x2b229fbd, 0x5174dca5, 0xf7028752, 0xd63c6aa8, 0x31276f38, 0x0646721c,
79     0xb0191da8, 0xe00e6de0, 0x9eac1a6e, 0x9f7628a5, 0xed6c06ea, 0x0bb8af15,
80     0xf119fb12, 0x38693c1c, 0x732bc0fe, 0x84953275, 0xb82ec888, 0x33a4f1b3,
81     0x3099835e, 0x028a8782, 0x5fdd51d7, 0xc6c717b3, 0xb06caf71, 0x17c8c111,
82     0x61bad754, 0x9fd03061, 0xe09df1af, 0x3bc9eb73, 0x85878413, 0x9889aaf2,
83     0x3f5a9e46, 0x42c9f01f, 0x9984a4f4, 0xd5de43cc, 0xd294daed, 0xbecba2d2,
84     0xf1f6e72c, 0x5551128a, 0x83af87e2, 0x6f0342ba,
85 };
86 
87 static uint64_t
88 hashblk(const unsigned char *p, off_t n)
89 {
90 	unsigned char buf[SHA1_DIGEST_LENGTH];
91 	uint64_t h;
92 	SHA1_CTX ctx;
93 	SHA1Init(&ctx);
94 	SHA1Update(&ctx, p, n);
95 	SHA1Final(buf, &ctx);
96 	memcpy(&h, buf, sizeof(h));
97 	return be64toh(h);
98 }
99 
100 static const struct got_error *
101 addblk(struct got_delta_table *dt, FILE *f, off_t file_offset0, off_t len,
102     off_t offset, uint64_t h)
103 {
104 	const struct got_error *err = NULL;
105 	int i;
106 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
107 	uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
108 	size_t r = 0;
109 
110 	if (len == 0)
111 		return NULL;
112 
113 	i = h % dt->nalloc;
114 	while (dt->blocks[i].len != 0) {
115 		/*
116 		 * Avoid adding duplicate blocks.
117 		 * NB: A matching hash is insufficient for detecting equality.
118 		 * The hash can only detect inequality.
119 		 */
120 		if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
121 			if (r == 0) {
122 				if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
123 					return got_error_from_errno("fseeko");
124 				r = fread(buf, 1, len, f);
125 				if (r != len) {
126 					if (!ferror(f))
127 						return NULL;
128 					return got_ferror(f, GOT_ERR_IO);
129 				}
130 			}
131 			if (fseeko(f, file_offset0 + dt->blocks[i].offset,
132 			    SEEK_SET) == -1)
133 				return got_error_from_errno("fseeko");
134 			if (fread(buf2, 1, len, f) != len)
135 				return got_ferror(f, GOT_ERR_IO);
136 			if (memcmp(buf, buf2, len) == 0)
137 				return NULL;
138 		}
139 
140 		i = (i + 1) % dt->nalloc;
141 	}
142 	assert(dt->blocks[i].len == 0);
143 	dt->blocks[i].len = len;
144 	dt->blocks[i].offset = offset;
145 	dt->blocks[i].hash = h;
146 	dt->nblocks++;
147 	if (dt->nalloc < dt->nblocks + 64) {
148 		struct got_delta_block *db;
149 		size_t old_size = dt->nalloc;
150 		db = dt->blocks;
151 		dt->blocks = calloc(dt->nalloc + 64,
152 		    sizeof(struct got_delta_block));
153 		if (dt->blocks == NULL) {
154 			err = got_error_from_errno("calloc");
155 			dt->blocks = db;
156 			return err;
157 		}
158 		dt->nalloc += 64;
159 		/*
160 		 * Recompute all block positions. Hash-based indices of blocks
161 		 * in the array depend on the allocated length of the array.
162 		 */
163 		dt->nblocks = 0;
164 		for (i = 0; i < old_size; i++) {
165 			if (db[i].len == 0)
166 				continue;
167 			err = addblk(dt, f, file_offset0, db[i].len,
168 			    db[i].offset, db[i].hash);
169 			if (err)
170 				break;
171 		}
172 		free(db);
173 	}
174 
175 	return err;
176 }
177 
178 static const struct got_error *
179 lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
180     unsigned char *p, off_t len, FILE *basefile, off_t basefile_offset0)
181 {
182 	int i;
183 	uint64_t h;
184 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
185 	size_t r;
186 
187 	*block = NULL;
188 
189 	h = hashblk(p, len);
190 	for (i = h % dt->nalloc; dt->blocks[i].len != 0;
191 	     i = (i + 1) % dt->nalloc) {
192 		if (dt->blocks[i].hash != h ||
193 		    dt->blocks[i].len != len)
194 			continue;
195 		if (fseeko(basefile, basefile_offset0 + dt->blocks[i].offset,
196 		    SEEK_SET) == -1)
197 			return got_error_from_errno("fseeko");
198 		r = fread(buf, 1, len, basefile);
199 		if (r != len)
200 			return got_ferror(basefile, GOT_ERR_IO);
201 		if (memcmp(p, buf, len) == 0) {
202 			*block = &dt->blocks[i];
203 			break;
204 		}
205 	}
206 	return NULL;
207 }
208 
209 static const struct got_error *
210 nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
211 {
212 	uint32_t gh;
213 	const unsigned char *p;
214 	size_t r;
215 	off_t pos = ftello(f);
216 
217 	*blocklen = 0;
218 
219 	r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
220 	if (r == 0 && ferror(f))
221 		return got_ferror(f, GOT_ERR_IO);
222 	if (r < GOT_DELTIFY_MINCHUNK)
223 		return NULL; /* no more delta-worthy blocks left */
224 
225 	/* Got a deltifiable block. Find the split-point where it ends. */
226 	p = buf + GOT_DELTIFY_MINCHUNK;
227 	gh = 0;
228 	while (p != buf + r) {
229 		gh = (gh << 1) + geartab[*p++];
230 		if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
231 			break;
232 	}
233 
234 	*blocklen = (p - buf);
235 	if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
236 		return got_error_from_errno("fseeko");
237 
238 	return NULL;
239 }
240 
241 const struct got_error *
242 got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
243     off_t filesize)
244 {
245 	const struct got_error *err = NULL;
246 	uint64_t h;
247 	const off_t offset0 = fileoffset;
248 
249 	*dt = calloc(1, sizeof(**dt));
250 	if (*dt == NULL)
251 		return got_error_from_errno("calloc");
252 
253 	(*dt)->nblocks = 0;
254 	(*dt)->nalloc = 128;
255 	(*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
256 	if ((*dt)->blocks == NULL) {
257 		err = got_error_from_errno("calloc");
258 		goto done;
259 	}
260 
261 	if (fseeko(f, fileoffset, SEEK_SET) == -1)
262 		return got_error_from_errno("fseeko");
263 
264 	while (fileoffset < filesize) {
265 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
266 		off_t blocklen;
267 		err = nextblk(buf, &blocklen, f);
268 		if (err)
269 			goto done;
270 		if (blocklen == 0)
271 			break;
272 		h = hashblk(buf, blocklen);
273 		err = addblk(*dt, f, offset0, blocklen,
274 		    fileoffset - offset0, h);
275 		if (err)
276 			goto done;
277 		fileoffset += blocklen;
278 		if (fseeko(f, fileoffset, SEEK_SET) == -1)
279 			return got_error_from_errno("fseeko");
280 	}
281 done:
282 	if (err) {
283 		free((*dt)->blocks);
284 		free(*dt);
285 		*dt = NULL;
286 	}
287 
288 	return err;
289 }
290 
291 void
292 got_deltify_free(struct got_delta_table *dt)
293 {
294 	if (dt == NULL)
295 		return;
296 	free(dt->blocks);
297 	free(dt);
298 }
299 
300 static const struct got_error *
301 emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
302     const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
303 {
304 	struct got_delta_instruction *d, *p;
305 
306 	if (*nalloc < *ndeltas + alloc_chunk_size) {
307 		p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
308 		    sizeof(struct got_delta_instruction));
309 		if (p == NULL)
310 			return got_error_from_errno("realloc");
311 		*deltas = p;
312 		*nalloc += alloc_chunk_size;
313 	}
314 	*ndeltas += 1;
315 	d = &(*deltas)[*ndeltas - 1];
316 	d->copy = copy;
317 	d->offset = offset;
318 	d->len = len;
319 	return NULL;
320 }
321 
322 static const struct got_error *
323 stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
324     FILE *f, off_t filesize, off_t *blocklen)
325 {
326 	uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
327 	size_t base_r, r, i;
328 	int buf_equal = 1;
329 
330 	if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
331 	    SEEK_SET) == -1)
332 		return got_error_from_errno("fseeko");
333 
334 	while (buf_equal && *blocklen < (1 << 24) - 1) {
335 		base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
336 		if (base_r == 0) {
337 			if (ferror(basefile))
338 				return got_ferror(basefile, GOT_ERR_IO);
339 			break;
340 		}
341 		r = fread(buf, 1, sizeof(buf), f);
342 		if (r == 0) {
343 			if (ferror(f))
344 				return got_ferror(f, GOT_ERR_IO);
345 			break;
346 		}
347 		for (i = 0; i < MIN(base_r, r); i++) {
348 			if (buf[i] != basebuf[i]) {
349 				buf_equal = 0;
350 				break;
351 			}
352 			(*blocklen)++;
353 		}
354 	}
355 
356 	return NULL;
357 }
358 
359 const struct got_error *
360 got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
361     FILE *f, off_t fileoffset, off_t filesize,
362     struct got_delta_table *dt, FILE *basefile,
363     off_t basefile_offset0, off_t basefile_size)
364 {
365 	const struct got_error *err = NULL;
366 	const off_t offset0 = fileoffset;
367 	size_t nalloc = 0;
368 	const size_t alloc_chunk_size = 64;
369 
370 	*deltas = NULL;
371 	*ndeltas = 0;
372 
373 	/*
374 	 * offset0 indicates where data to be deltified begins.
375 	 * For example, we want to avoid deltifying a Git object header at
376 	 * the beginning of the file.
377 	 */
378 	if (fseeko(f, offset0, SEEK_SET) == -1)
379 		return got_error_from_errno("fseeko");
380 
381 	*deltas = reallocarray(NULL, alloc_chunk_size,
382 	    sizeof(struct got_delta_instruction));
383 	if (*deltas == NULL)
384 		return got_error_from_errno("reallocarray");
385 	nalloc = alloc_chunk_size;
386 
387 	while (fileoffset < filesize) {
388 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
389 		off_t blocklen;
390 		struct got_delta_block *block;
391 		err = nextblk(buf, &blocklen, f);
392 		if (err)
393 			break;
394 		if (blocklen == 0) {
395 			/* Source remainder from the file itself. */
396 			if (fileoffset < filesize) {
397 				err = emitdelta(deltas, &nalloc, ndeltas,
398 				    alloc_chunk_size, 0, fileoffset - offset0,
399 				    filesize - fileoffset);
400 			}
401 			break;
402 		}
403 		err = lookupblk(&block, dt, buf, blocklen, basefile,
404 		    basefile_offset0);
405 		if (err)
406 			break;
407 		if (block != NULL) {
408 			/*
409 			 * We have found a matching block in the delta base.
410 			 * Attempt to stretch the block as far as possible and
411 			 * generate a copy instruction.
412 			 */
413 			err = stretchblk(basefile, basefile_offset0, block,
414 			    f, filesize, &blocklen);
415 			if (err)
416 				break;
417 			err = emitdelta(deltas, &nalloc, ndeltas,
418 			    alloc_chunk_size, 1, block->offset, blocklen);
419 			if (err)
420 				break;
421 		} else {
422 			/*
423 			 * No match.
424 			 * This block needs to be sourced from the file itself.
425 			 */
426 			err = emitdelta(deltas, &nalloc, ndeltas,
427 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
428 			if (err)
429 				break;
430 		}
431 		fileoffset += blocklen;
432 		if (fseeko(f, fileoffset, SEEK_SET) == -1) {
433 			err = got_error_from_errno("fseeko");
434 			break;
435 		}
436 	}
437 
438 	if (err) {
439 		free(*deltas);
440 		*deltas = NULL;
441 		*ndeltas = 0;
442 	}
443 	return err;
444 }